Enhancement of the Computational Performance for Near- and Far-Field Target Strength Analysis Using OptiX-based Shooting and Bouncing Rays Method

Target strength (TS) is one of the critical design factors for evaluating the susceptibility of naval ship. Extensive researches have been accomplished to analyze and estimate TS, and software developed based on these researches provides highly accurate solutions for arbitrary shapes. Recently, as submarines become larger and signature reduction technologies advance rapidly in various ways, the computational burden of TS analysis has increased. With the rapid progress of graphic processing unit (GPU) technology optimized for high speed computation, there has been growing the interest in improving the computational efficiency of TS analysis. However, most studies aimed at improving the computational performance of TS analysis assume far-field conditions in which the distance between the sonar and the target is sufficiently far, leading to limitations when applied to near-field acoustic problems where far-field conditions are not satisfied. In this study, a GPU-based TS analysis methodology applicable to both far- and near-field conditions within the cases considered in this study is proposed using the shooting and bouncing rays (SBR) method. The entire process of TS analysis is parallelized on the GPU by utilizing NVIDIA OptiX, a high-performance ray-tracing library. A series of TS analyses are performed using simple-shaped geometries, including a plate, sphere, and dihedral, as well as the benchmark target strength simulation (BeTSSi) model to validate its applicability and effectiveness. In addition, the comparison with conventional CPU-based TS analysis confirms that the proposed GPU-based approach provides considerable improvements in computational efficiency while maintaining reliable accuracy.